# Will the US Bureau of Labor Statistics count fewer than 2.6 million customer service representatives in its May 2027 job data?

> As of 17 November 2026, Prolepsia puts the chance of yes at 92%.

- Forecaster: Prolepsia (https://prolepsia.com)
- Forecast made: 2026-11-17
- Question type: Yes / no
- Resolves: 2027-05-31
- Status: Open
- Forecasts: 1
- Page: https://prolepsia.com/forecasts/equinox-lattice-pulsar/will-the-us-bureau-of-labor-statistics-count-fewer-than-2-6-million-customer-ser
- Open in the app: https://app.prolepsia.com/library/320074a0-c9b1-4ec2-9fb3-483eb3bb82fd

## Forecast

- Yes: 92%
- No: 8%

## How Prolepsia reasons

## TL;DR
The forecast is **92% Yes and 8% No** for the first published May 2027 OEWS national employment estimate for customer service representatives falling strictly below 2,600,000. Two substantial annual declines and continued call-center payroll losses support Yes. The decisive issue is whether that decline persists, not whether AI can perform customer-service tasks.

## Context
The question concerns one occupation, one statistical program, and one reference month: Customer Service Representatives, SOC 43-4051, in the national cross-industry OEWS estimate for May 2027. OEWS covers wage-and-salary employment in nonfarm establishments and excludes self-employed workers and other specified categories. It does not count all customer-facing workers, vacancies, or full-time-equivalent positions. ([BLS program overview](https://www.bls.gov/oes/oes_emp.htm))

As of November 17, 2026, the latest qualifying observation is May 2025. BLS published that estimate on May 15, 2026, reporting 2,595,750 jobs. ([BLS release](https://www.bls.gov/news.release/archives/ocwage_05152026.htm)) The question will use the first published May 2027 estimate, expected in 2028, without later revisions. An estimate of exactly 2,600,000 resolves No; only a lower count resolves Yes.

## Evidence
The historical record is the backbone of the forecast. This occupation has experienced both sustained growth and contraction. A forecast based only on the latest two declines would miss that history. The national OEWS employment counts are shown below. ([BLS historical tables](https://www.bls.gov/oes/tables.htm))

| May reference year | National employment |
|---|---:|
| 2010 | 2,146,120 |
| 2011 | 2,212,820 |
| 2012 | 2,299,750 |
| 2013 | 2,389,580 |
| 2014 | 2,511,130 |
| 2015 | 2,595,990 |
| 2016 | 2,707,040 |
| 2017 | 2,767,790 |
| 2018 | 2,871,400 |
| 2019 | 2,919,230 |
| 2020 | 2,833,250 |
| 2021 | 2,787,070 |
| 2022 | 2,879,840 |
| 2023 | 2,858,710 |
| 2024 | 2,725,930 |
| 2025 | 2,595,750 |

The record shows a long expansion through 2019, declines in 2020 and 2021, and a rebound in 2022. Employment then fell in each of the next three observations. The latest two losses were particularly large: 132,780 jobs followed by 130,180 jobs. The cumulative May 2023–May 2025 reduction was 262,960 jobs, or 9.1986%. This repeated contraction carries more weight than a single below-threshold observation. ([BLS historical tables](https://www.bls.gov/oes/tables.htm))

The starting margin is nevertheless tiny. The May 2025 estimate sits only 4,250 jobs below the threshold. Its reported employment relative standard error is 0.4%, implying a sampling standard error of 10,383 jobs. The margin is smaller than that sampling-error measure. This does not establish the chance of a future threshold crossing, but it shows why “already below” is not enough on its own. ([BLS May 2025 national data](https://www.bls.gov/oes/special-requests/oesm25nat.zip))

More recent payroll data supports continued weakness in specialized service providers. The following seasonally adjusted observations cover contact centers and the broader telephone-call-center industry. They are industry employment figures across all occupations, not national customer service representative counts. September 2026 is preliminary. ([BLS payroll table](https://www.bls.gov/ces/data/employment-and-earnings/2026/table1a_202610.htm))

| Reference month | Contact centers | Telephone call centers |
|---|---:|---:|
| May 2025 | 297,800 | 331,800 |
| September 2025 | 287,900 | 322,200 |
| December 2025 | 283,100 | 317,500 |
| January 2026 | 279,100 | 312,200 |
| February 2026 | 279,000 | 311,100 |
| March 2026 | 278,800 | 311,000 |
| April 2026 | 277,800 | 309,800 |
| May 2026 | 276,400 | 307,800 |
| June 2026 | 277,500 | 309,100 |
| July 2026 | 278,200 | 308,900 |
| August 2026 | 277,700 | 308,200 |
| September 2026 | 276,200 | 307,000 |

The direction is clear, but the pace has slowed. Telephone-call-center employment fell substantially between May 2025 and May 2026, then changed much less through September. Contact-center employment was also nearly flat over that later interval. This supports continued pressure, not a mechanical extension of the earlier rate of decline. The two industries overlap, so their losses cannot be added. ([BLS payroll table](https://www.bls.gov/ces/data/employment-and-earnings/2026/table1a_202610.htm))

The household survey offers separate directional support. Its annual-average customer service representative employment estimates were 2,822,000 in 2023, 2,632,000 in 2024, and 2,600,000 in 2025. The final observation averages only 11 months because October data were not collected during the shutdown. These figures measure a different employment universe and are not substitutes for OEWS. They also suggest a slowing decline rather than accelerating contraction. ([BLS household-survey table](https://www.bls.gov/cps/cpsaat11.htm))

Hiring demand provides a counter-signal. Indeed's customer-service postings index rose from a monthly average of 81.9575 in May 2025 to 86.9697 in October 2026. The latest reported seven-day-average observation was 87.46 on November 6, 2026. All remained below the February 1, 2020 baseline of 100. A recovering stock of postings can precede stronger hiring, but it does not establish growth in employment or precisely match SOC 43-4051. ([Indeed/FRED series](https://fred.stlouisfed.org/series/IHLIDXUSTPCUSTSERV))

BLS's structural outlook favors decline at a much slower pace than the latest OEWS losses. Its separate employment-projections program puts employment at 2,666,000 in 2025 and projects a 5.3% decline through 2035. Automation and self-service reduce demand for routine work, while complex inquiries continue to require people. This supports the direction of the forecast without making the projection's employment level interchangeable with OEWS. ([BLS Occupational Outlook Handbook](https://www.bls.gov/ooh/office-and-administrative-support/customer-service-representatives.htm))

The AI evidence is mixed. A study of 5,172 customer-support agents found a 15.2% productivity improvement, with larger gains among novices. Most agents were outside the United States. The result shows that assistance can raise output per worker; it does not establish an equivalent reduction in national employment. ([Generative AI at Work](https://oup.silverchair-cdn.com/article-minimal/7990658)) A randomized experiment at Alibaba also found service-performance gains, but it studied agents in China and did not measure national employment displacement. ([Alibaba study](https://arxiv.org/pdf/2603.29888))

Employer evidence shows both substitution and limits. Gartner's November 5, 2026 survey reported that 21% of service and support leaders had reduced human-agent headcount because of AI. It also found that 7% had rehired and 11% planned rehiring. The survey was not established as U.S.-representative or employment-weighted, so those responses cannot be converted into a national job-loss rate. ([Gartner survey](https://www.gartner.com/en/newsroom/press-releases/2026-11-05-gartner-survey-finds-21-percent-of-service-and-support-leaders-have-reduced-human-agent-headcount-because-of-ai))

Other research cuts against a uniform displacement story. Stanford/ADP reports weaker customer-service employment among younger workers but expansion among older groups relative to November 2022. ([Stanford dashboard](https://digitaleconomy.stanford.edu/project/indicators/canaries-dashboard/)) Ramp/Revelio found customer-service headcount growth among high-intensity AI adopters, though employer selection and faster growth among adopters limit causal interpretation. ([Ramp/Revelio paper](https://ramp.com/data/ai-jobs-impact/paper)) A Danish study found limited early effects on hours and earnings, but its geography and earlier observation period limit its relevance to this U.S. outcome. ([Danish study](https://www.nber.org/system/files/working_papers/w33777/w33777.pdf))

Domestic hiring commitments are another offset. AT&T announced that specified consumer services would be handled by U.S.-located Care experts by end-2027. It did not specify a customer-service job increment, and the completion date falls after the target reference month. The announcement supports a possible offset, not a quantified increase by May 2027. ([AT&T announcement](https://www.attconnects.com/stories/not-all-good-jobs-require-a-college-degree/))

## What's non-obvious
**This is not a bet on an AI-driven collapse.** The threshold sits just above the latest OEWS estimate. A Yes outcome does not require another decline of the recent size, or even a decline from May 2025. The strong Yes case rests on sustained occupational weakness and subsequent industry contraction. AI provides a plausible mechanism, but the observed employment path does more work than claims about technical capability.

The other easy mistake is to treat call centers as the whole occupation. Customer service representatives also work inside retailers, banks, insurers, and other organizations. A specialized provider can shrink because work moves in-house, not because the occupation loses the same number of jobs nationally. And reduced junior hiring can coexist with retained or growing senior employment. The evidence therefore supports contraction without supporting a one-for-one conversion from AI adoption or call-center losses to nationwide occupational losses.

## Uncertainties
The main risk is stabilization or a rebound. Recovering postings, retained human agents, domestic hiring, and demand outside specialized providers can all push against decline. Because the initial margin is only 4,250 jobs, a small increase from the latest OEWS level is enough for No. More current, nationally representative occupational staffing data outside call centers would most directly clarify this risk.

Survey construction adds uncertainty. OEWS combines six semiannual panels over three years and benchmarks establishment employment. Staffing changes enter gradually, while panel turnover and estimation affect the published level. BLS cautions against straightforward time-series comparisons, and the May 2021 methodological change weakens comparisons across the full historical record. Sampling error captures only part of this uncertainty. ([BLS technical notes](https://www.bls.gov/oes/2025/may/oes_tec.htm); [BLS time-series cautions](https://www.bls.gov/oes/oes_ques.htm))

The exact national employment effect of AI remains unverified. Productivity studies, employer surveys, age-specific payroll patterns, and company announcements answer different questions. More representative evidence linking deployments to U.S. occupational hiring, departures, and headcount would close that gap. These uncertainties leave room for No, but they do not overturn the forecast's central conclusion: the first published May 2027 OEWS estimate will fall below the threshold.

## Forecast history

| Date | Forecast |
|---|---|
| 2026-11-17 | 92% chance of yes |

## Sources Prolepsia read

- [Occupational Employment and Wage Statistics Overview : U.S. Bureau of Labor Statistics](https://www.bls.gov/oes/oes_emp.htm)
- [Customer Service Job Postings on Indeed in the United States (IHLIDXUSTPCUSTSERV) \| FRED \| St. Louis Fed](https://fred.stlouisfed.org/series/IHLIDXUSTPCUSTSERV)
- [Generative AI at Work](https://oup.silverchair-cdn.com/article-minimal/7990658)
- [Alibaba study](https://arxiv.org/pdf/2603.29888)
- [Gartner Survey Finds 21% of Service and Support Leaders Have Reduced Human Agent Headcount Because of AI](https://www.gartner.com/en/newsroom/press-releases/2026-11-05-gartner-survey-finds-21-percent-of-service-and-support-leaders-have-reduced-human-agent-headcount-because-of-ai)
- [Canaries Dashboard - Stanford Digital Economy Lab](https://digitaleconomy.stanford.edu/project/indicators/canaries-dashboard)
- [Ramp/Revelio paper](https://ramp.com/data/ai-jobs-impact/paper)
- [Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AI](https://www.nber.org/system/files/working_papers/w33777/w33777.pdf)
- [Not All Good Jobs Require A College Degree - AT&T Connects](https://www.attconnects.com/stories/not-all-good-jobs-require-a-college-degree)
- [Canaries Dashboard: Employment in AI-exposed occupations shrank in September](https://www.adpresearch.com/research/canaries-dashboard-employment-in-ai-exposed-occupations-shrank-in-september)

## The question

This question asks whether the U.S. Bureau of Labor Statistics (BLS) will estimate that fewer than 2,600,000 people were employed as Customer Service Representatives (Standard Occupational Classification code 43-4051) nationwide in May 2027, according to its Occupational Employment and Wage Statistics (OEWS) program. The question is being posed on November 17, 2026, and will be resolved after the BLS publishes its May 2027 OEWS estimates, expected in 2028. The OEWS program publishes annual occupational employment estimates covering wage and salary workers in nonfarm establishments across the United States. It excludes self-employed workers, certain business owners, unpaid family workers, and private household workers. The latest available OEWS data, for May 2025, estimated 2,595,750 Customer Service Representatives nationwide, already slightly below the proposed threshold of 2.6 million. This represents a decline from the approximately 2.86 million estimated in May 2023. Separately, the BLS Occupational Outlook Handbook estimates 2,666,000 Customer Service Representative jobs in 2025 using its National Employment Matrix methodology and projects employment in this occupation will decline by approximately 5.3% between 2025 and 2035, partly because of increased automation, self-service systems, and technological improvements. These figures come from different BLS statistical programs and are not directly interchangeable. This question specifically concerns whether the May 2027 OEWS national employment estimate will fall below 2.6 million, rather than whether the BLS projects a decline or whether employment falls below that threshold at any other point in time. Sources: BLS May 2025 Occupational Employment and Wages (https://www.bls.gov/news.release/ocwage.htm), BLS Occupational Employment and Wage Statistics Overview (https://www.bls.gov/oes/oes_emp.htm), and BLS Customer Service Representatives Occupational Outlook Handbook (https://www.bls.gov/ooh/office-and-administrative-support/customer-service-representatives.htm).

### How it resolves

The question resolves YES if the U.S. Bureau of Labor Statistics reports a national employment estimate strictly below 2,600,000 for Customer Service Representatives (SOC code 43-4051) in its official May 2027 Occupational Employment and Wage Statistics (OEWS) dataset. The question resolves NO if the corresponding estimate is 2,600,000 or greater. The primary resolution source will be the BLS May 2027 OEWS national occupational employment estimates, specifically the employment figure for occupation 43-4051 in the national cross-industry data. These estimates should be available through the BLS OEWS website (https://www.bls.gov/oes/) after publication, normally in 2028. The precise national employment count published by BLS will be used, rather than a rounded figure expressed in millions. The outcome will be determined using the first officially published May 2027 OEWS estimate, without incorporating subsequent revisions. If BLS publishes the May 2027 estimates but omits this occupation, the resolution will be deferred while BLS is contacted or additional official documentation is checked. If the occupation is permanently discontinued, merged with other occupations, or otherwise made unavailable without a directly comparable national estimate, the question will be annulled. If the May 2027 OEWS dataset is not published by December 31, 2029, the question will also be annulled rather than automatically resolving NO. If BLS changes its occupational classification system but explicitly provides a directly comparable employment estimate for Customer Service Representatives, that official estimate will be used. Estimates from other BLS programs, including the Occupational Outlook Handbook, National Employment Matrix, Current Population Survey, and Current Employment Statistics, will not substitute for the OEWS estimate.

### Fine print

The threshold is strictly fewer than 2,600,000 employed Customer Service Representatives. An estimate of exactly 2,600,000 resolves NO, while an estimate of 2,599,990 resolves YES. The geographic scope is the United States as represented by the BLS national OEWS employment estimates, using the geographic coverage defined by that program. The occupation is defined by SOC code 43-4051, not by a broader category of customer-facing employees or all workers involved in customer support. Workers in sales, technical support, management, or other occupations are included only if BLS classifies their primary duties under SOC 43-4051. The measure is the BLS estimate of employment, not the number of job vacancies, job postings, full-time-equivalent positions, or projected future jobs. May 2027 refers to the reference period of the employment estimates, not the publication month. A May 2027 dataset released in 2028 will therefore qualify. The question does not require employment to remain below 2.6 million for any minimum duration. Because the May 2025 OEWS estimate was already slightly below 2.6 million, a YES outcome would not necessarily indicate a further decline from the most recent available estimate.

---

Prolepsia’s forecasting engine made this forecast: it researched the question, weighed the evidence and wrote the report on this page. A probability is not a promise; this one changes as the evidence does.
